soongsc said:
Thanks, Soongsc
I'd like to add being very grateful to Earl for his huge interest in this topic and for playing the part of "advocatus diaboli" – a very positive and highly necessary part of any good "in depth" discussion (though pretty destructive if one gets stuck in that role)
After all, it's been Earl who has triggered this thread by his request and interest – having speeded things (my deeper investigation of TD on the level of wave form deformation) up reasonable, by demanding "the full package".
Pan said:
(hang on, i will contradict myself below..
I love that too !
soongsc said:
All *I* can say is that TD adds kind of sonic signature – or better - different sonic signatures on different time scales.
soongsc said:
Can't follow you ??
soongsc said:Pan said:
What you do is very interesting though and I was thinking on the constant voltage vs. constant power thing. To me it makes sense to measure (and simulate) with constant voltage since most amp/speaker interfaces is about voltage drive. Or am I thinking wrong here?
(hang on, i will contradict myself below..
Typically amp source impedance and cable impedance is out of the equation but the crossover isn't and that is an interesting thing.
Power compression is worst when there is a voltage drive and low impedance between the amp output and the voice coil, but what happens when we put coils and resistors in series with a driver is that the effect of the heating is decreased. Also the HD/IMD is typically decreased with a coil in series with a speaker driver, at least if there's significant inductance in the area of interest.
I have not measured these effects myself but they are there and others have looked into it.
To sum up the confusion above, examine a raw driver would be best to use constant voltage even though the driver will be used with a passive x-over which skew things towards current drive.
Pan said:
Yes the source impedance comes into play - be it the crossover, amp, cable or whatever..
I have outlined this in short in my paper for the trade off between voltage amp and current amp.
There are two things to be considered.
- First - at what absolute value is the source resistance with respect to the VC impedance. To be more precise – its *only the resistive* part of that impedance that is affected by Thermal Distortion.
- second – what are the temperature coefficients in relation to each other
You can look at it – and could measure it as well - like a Whetstone bridge – a concept many people are familiar with.
As for the "dialectic dialogue" of power versus voltage
As for me, I try to keep it as simple and intuitively understandable as possible – hence its a no-brainer to continue with measuring at constant voltage and call it constant power injection.
It's much easier to perform measurements – no small concern if you would like to design a standardisation concept - and its easy to correct deviations by math and simulation.- *if* one is interested in *absolutely correct* figures with an infinite number of decimal places
Michael
Is it possible to describe what kind of sonic signature? In what kind of music is it more evident?mige0 said:
Well, it seems that thermal distortion is related with varying of temperature. In real music situation, since the power level is varying, the temperature will not vary that much meaning, with the additional heat dissipation capability when the vc is in the gap, the distortion levels would be even lower. Probably there is a test that will reveal the difference between bare vc and vc in gap of well designed motor.
soongsc said:
At the moment this is asked for too much – I'm afraid.
At least in an more pronounced and pinpointed fashion as many already have expressed. We simply don't have enough drivers measured for Thermal Distortion to do that.
It just had happened that we gained some deeper understanding on the underlying mechanisms of Thermal Distortion – which hopefully will spread out into common knowledge among audio enthusiasts..
Until now Thermal Distortion is only characterised in its "power compression" facet – which in fact is only one out of three major effects – as outlined earlier.
This has strong historic roots in the PRO bizz – not so easy to overcome.
Now that we *think* we know better (*if* my thermal model gets validated) and now that we (possibly) were able to tell the whole quality of TD (with simplifications and limitations) – now that the thermal effects got isolated and clearly enough could have been described and defined what can be subsumed under Thermal Distortion and what not - I can see basically two possible ways to come to a more "serious" definition about audible impacts of Thermal Distortion.
(Ups – what a sentence...)
One way is the classical scientific approach of doing controlled listening tests – Earl already has outlined this path, and, hopefully will go this route
The other way (the slower one) is the classical comparative method – which anyway happens when something "new" calls our attraction - on a less scientific level.
First we will be tempted to quantify the effect in a more or less standardised technical form. (I already have made suggestions for that and may come back to it more in detail). The current "power compression" figures are of little help as they are pretty meaningless in a more holistic context.
Once there is a large enough TD database covering a diverse set of speakers – anybody can – and will - draw conclusions from his own experience with different driver topologies.
Just think of how different ribbons are to dynamic speakers or compared to electrostatic speakers with respect to Thermal Distortion.
My private guess is that scientific method will reveal that there is no audible correlation whatsoever – no wonder as pretty no one has stumbled over this the last 100 years or so – meaning its a subtle effect 99.99% of listener's won't care – and also no patents will offer any additional income and turn on any big advertising machinery.
Those ones like Wayne Parham, AE-Speakers, Brian Ding to name a few appearing around here who were into this very kind of thing on a manufacturer level – didn't made it into *common* consciousness with their ideas and products until now.
But anyway - for those demanding top notch quality – TD will become an issue more and more clarified in terms of audibility over time.
After all its been the descriptions of a "feeling" about a uncertain something that lead to what I investigated.
What we gained from my work is only "a good first step"
What I outlined above also makes your work / study very interesting for me - Peter !
Maybe we could find a common basis about sort of standardisation of the measurement procedure in order to make our results and findings comparable - once published ?
Michael
mige0 said:
Michael
I am not so sure about this. What you have not done is to modulate the signal in a realistic way so as to see the heating and cooling effects overlaped. I do not believe that the heating and cooling are symmetric even though your model shows this. I think that in the real world they will be quite different and this will have significant implications for real world dynamics.
But we will see.
Maybe this week I can make that modulated noise signal. I've though about it and now know how to do it and thats a big step.
Earl, "symmetric" might have been somewhat misleading.
What I meant is that the TD amplitude of 50% duty cycle bursts are symmetric to its mean value – at thermal equilibrium – not that the shape of the envelope is symmetric.
The shape of the amplitude envelope is the well known trumpet and its inverse on SPL melt down and SPL recovery respectively.
Here are the pix I showed to illustrate this:
From measurement posting #44:
http://www.diyaudio.com/forums/showthread.php?postid=1667982#post1667982
From simulation #32:
http://www.diyaudio.com/forums/showthread.php?postid=1666398#post1666398
From simulation #60:
http://www.diyaudio.com/forums/showthread.php?postid=1669659#post1669659
The explanation for this:
"And also remember - once the motor is hot (at equilibrium to be precise) its exactly the stage where symmetric heating and cooling happen - to maintain that constant equilibrium temperature of the motor."
was also outlined for Nick at #44
Michael
What I meant is that the TD amplitude of 50% duty cycle bursts are symmetric to its mean value – at thermal equilibrium – not that the shape of the envelope is symmetric.
The shape of the amplitude envelope is the well known trumpet and its inverse on SPL melt down and SPL recovery respectively.
Here are the pix I showed to illustrate this:
From measurement posting #44:
http://www.diyaudio.com/forums/showthread.php?postid=1667982#post1667982
An externally hosted image should be here but it was not working when we last tested it.
From simulation #32:
http://www.diyaudio.com/forums/showthread.php?postid=1666398#post1666398
An externally hosted image should be here but it was not working when we last tested it.
From simulation #60:
http://www.diyaudio.com/forums/showthread.php?postid=1669659#post1669659
An externally hosted image should be here but it was not working when we last tested it.
The explanation for this:
"And also remember - once the motor is hot (at equilibrium to be precise) its exactly the stage where symmetric heating and cooling happen - to maintain that constant equilibrium temperature of the motor."
was also outlined for Nick at #44
Michael
If you go back to #51
http://www.diyaudio.com/forums/showthread.php?postid=1668717#post1668717
Wayne Parham describes this effect form his own experience with bass speakers.
As – what I call - "thermal overshot" is a function of the absolute SPL meltdown at a given power injection *and* the frequency of the 50% dutycycle with respect to the VC + motor time constant – we could expect the speaker Wayne was listening to had either a severe absolute SPL meltdown at this power injection and/or its time constant was in the range (or less) of the 15 sec mentioned.
Michael
http://www.diyaudio.com/forums/showthread.php?postid=1668717#post1668717
Wayne Parham describes this effect form his own experience with bass speakers.
As – what I call - "thermal overshot" is a function of the absolute SPL meltdown at a given power injection *and* the frequency of the 50% dutycycle with respect to the VC + motor time constant – we could expect the speaker Wayne was listening to had either a severe absolute SPL meltdown at this power injection and/or its time constant was in the range (or less) of the 15 sec mentioned.
Michael
Michael
Certainly once a system reaches thermal equilibrium the heating and cooling must be equal and symmetric. But I agree with Nick, this is never going to happen in reality, or its going to take a very long time. But for a dynamic signal there is never going to be a steady state and as such there is never going to be equal heating and cooling effects. To understand the problem, I believe that we must look at this effect.
Certainly once a system reaches thermal equilibrium the heating and cooling must be equal and symmetric. But I agree with Nick, this is never going to happen in reality, or its going to take a very long time. But for a dynamic signal there is never going to be a steady state and as such there is never going to be equal heating and cooling effects. To understand the problem, I believe that we must look at this effect.
YES – and – NO...
Thermal equilibrium is never going to be happen – or – thermal equilibrium is happening pretty quick.
Both is right – depending on how close we look at the errors remaining.
Equilibrium is reached to 99% after only (!) 5 times the time constant of our voice coil (in case the motor time constant does not play a dominant role)
This is – for the compression driver presented - after a very short 10- 20 seconds.
Once thermal equilibrium is reached it *only* depends on the time constant with respect to the duty cycle frequency how wide "thermal overshot" swings around the mean value of thermal melt down.
This is not really intuitively – if one comes fresh to the topic.
Of course "thermal overshot" can't swing *more* in amplitude than the mean value. In other words the VC can't cool down deeper as ambient temperature allows and where we were starting from.
*If* we make the thermal mass very very very small, it does not take any measurable time to heat up the VC. – meaning – it would heat up immediately to the temperature where cooling and heating due to power injection have its equilibrium (remember: IMMEDIATELY - *if* no thermal mass were involved)
If we set duty cycle to 50% - the same time span VC temperature IMMEDIATELY would reach (and stay at ) ambient temperature.
Hence – the mean value would be at halve between and the shape would be that of a square signal rather than looking like Willy's fin
Instead of making the mass very very very small – which is not possible in real world – we also could make duty cycle frequency very very very low instead.
The woofer Wayne was auditioning was close enough to equilibrium after only a few minutes of heat up to make the more or less symmetric melt down and recovery clearly audible.
Above is valid for any "steady" duty cycled signals – which of course is somewhat different to any music program.
*But* - I can't see any additional effects not yet outlined though, regarding music as stimulus. Maybe I haven't fully understood what you mean ?
All is covered by the same simulation (within its outlined limits)
You can feed my thermal model basically with any signal you like – as shown for the mixed two tones for example
But - lets see what additional insights we can get from your specially composed stimulus file
Michael
Thermal equilibrium is never going to be happen – or – thermal equilibrium is happening pretty quick.
Both is right – depending on how close we look at the errors remaining.
Equilibrium is reached to 99% after only (!) 5 times the time constant of our voice coil (in case the motor time constant does not play a dominant role)
This is – for the compression driver presented - after a very short 10- 20 seconds.
Once thermal equilibrium is reached it *only* depends on the time constant with respect to the duty cycle frequency how wide "thermal overshot" swings around the mean value of thermal melt down.
This is not really intuitively – if one comes fresh to the topic.
Of course "thermal overshot" can't swing *more* in amplitude than the mean value. In other words the VC can't cool down deeper as ambient temperature allows and where we were starting from.
*If* we make the thermal mass very very very small, it does not take any measurable time to heat up the VC. – meaning – it would heat up immediately to the temperature where cooling and heating due to power injection have its equilibrium (remember: IMMEDIATELY - *if* no thermal mass were involved)
If we set duty cycle to 50% - the same time span VC temperature IMMEDIATELY would reach (and stay at ) ambient temperature.
Hence – the mean value would be at halve between and the shape would be that of a square signal rather than looking like Willy's fin
An externally hosted image should be here but it was not working when we last tested it.
Instead of making the mass very very very small – which is not possible in real world – we also could make duty cycle frequency very very very low instead.
The woofer Wayne was auditioning was close enough to equilibrium after only a few minutes of heat up to make the more or less symmetric melt down and recovery clearly audible.
Above is valid for any "steady" duty cycled signals – which of course is somewhat different to any music program.
*But* - I can't see any additional effects not yet outlined though, regarding music as stimulus. Maybe I haven't fully understood what you mean ?
All is covered by the same simulation (within its outlined limits)
You can feed my thermal model basically with any signal you like – as shown for the mixed two tones for example
But - lets see what additional insights we can get from your specially composed stimulus file
Michael
gedlee said:
Maybe it helps to look at this from an other perspective.
Temperature rise and temperature fall are a *consequence* of the imbalance of power injection and power loss (heating and cooling)
Exactly what we are interested in – no?
Its the mental hurdle of thermal mass that complicate things.
Take out of the equation thermal mass (as a mental experiment) and things pretty soon become crystal clear.
Michael
Michael,
I'll be glad to discuss these matters and share ideas with you.
There's a lot of frequency and distortion measuring going on but few (in the open and in the DIY arena) seems to have put effort in thermal compression.
Of course there are people that have been studying this in depth but little seems to have been published.
/Peter
I'll be glad to discuss these matters and share ideas with you.
There's a lot of frequency and distortion measuring going on but few (in the open and in the DIY arena) seems to have put effort in thermal compression.
Of course there are people that have been studying this in depth but little seems to have been published.
/Peter
Michael
The problem that I have with your model is that it is linear. The resistors, which represent heat flow, are linear and heat flow is not. This model is OK for some effects, but not in detail.
In fact, I am a little concerned with how the correlation approach that I want to use will work with nonlinear aspects of the problem. Correlation is a linear concept.
The problem that I have with your model is that it is linear. The resistors, which represent heat flow, are linear and heat flow is not. This model is OK for some effects, but not in detail.
In fact, I am a little concerned with how the correlation approach that I want to use will work with nonlinear aspects of the problem. Correlation is a linear concept.
gedlee said:
Hi Earl, yes, I'm with you on that. Hence the question is, when does the Re's greatest rate of change occur as this where the effect(s) will be greatest? At the beginning?, somewhere in the middle?, obviously not when settled at high levels (I would assume a fair bit of heat stored). I think stability of Re is the key here, whatever the 'magnitude' (sensitivity to audibility - I may have some ideas on that) that John mentioned earlier.
For fear of stating the obvious, it's not the temperature perse that causes the problem, the driver already starts at a static point. When it becomes closer to static, albeit at a higher temp, then eroded Qe and lower sens, but compression becomes lower.
In my mind I come back to your earliest points re thermal mass. While we figure that this may be a slow process and looking like out of band, yet we seem to say that the change is audible (agree with his comment re larger v. small - Post #52), so we need to look at where the greatest rate of change occurs, irrespective of its equivalent frequency, because without it we shouldn't hear anything wrong.
I have some additional thoughts on this, some of which enters a more psycho-acoustic idea. Anybody ever thought of the effect of pre-heating VC and keeping it there to lower rate of change? Would that work? How about a thermal servo? I'm kidding, I think.
Joe R.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.